CN101564289A - Method for real-time error correction of neurosurgery navigation puncture path based on near infrared spectrum - Google Patents

Abstract

The invention discloses a near-infrared spectrum-based neurosurgery navigation puncture path real-time error correction method, which belongs to a surgical navigation system precision correction method. The present invention first uses the volume data based on the three-dimensional reconstruction volume to extract the gray level information of the target path. Then the target path is conicalized, and the gray information of N verification puncture trajectories is extracted on the cone surface. The optical curves of the target path and N verification trajectories are obtained from the mathematical correlation model of optical parameters and image data. With the help of the curve trend matching algorithm of Hausdorff distance and curvature, the same variation trend as the real-time near-infrared optical signal is obtained within the tolerance range. The puncture trajectory curve can be used to determine the orientation that deviates from the target path, calculate the coordinate offset, perform real-time correction of the path, and guide the operation. The method is simple to implement, fast in speed, can be performed in real time, is convenient for clinical application, and can be integrated in a surgical navigation system, thereby greatly improving the accuracy of the navigation system.

Description

Method for real-time error correction of neurosurgery navigation puncture path based near infrared spectrum

Technical field

Invention belongs to Medical Image Processing, Biomedical Photonics and application, relates to a kind of surgical navigation systems accuracy correcting method, is specifically related to a kind of method for real-time error correction of neurosurgery navigation puncture path based near infrared spectrum

Background technology

In the neurosurgery navigation system, can't proofread and correct the release because of cerebrospinal fluid, the cerebral tissue displacement that cerebral tissue tractive, Body Position Change etc. cause is the key factor of surgical navigation systems precision that affects the nerves.The actual puncture path how correction cerebral tissue displacement in real time brings in operation process and the problem of surgical planning path offset are the clinical medicine problems that the neurosurgery navigation system perplexs for a long time and faces the challenge.Solving the ideal method of cerebral tissue displacement at present is the neural navigating surgery that adopts the guiding of low exploitation formula nuclear magnetic resonance, NMR, but because operation needs complete armoured magnetic field, operating theater instruments and microscope are extraordinary demagnetization material, the operation cost is too high, the expense costliness, complicated operation is difficult to penetration and promotion.The major technique of clinical practice at present is the discharge of microelectrode recording cell] and ultrasonic assisting navigation.Microelectrode recording cell discharge characteristic, can reach and dissect upward and the dual location on the function, effectively avoid damaging target spot important structure such as capsula interna, tractus opticus on every side, shortcoming is that the clinicist relies on experience decision signal feature, there is not quantitative identification parameter, the judgement that relies on sudden strain of a muscle screen and sound to make lacks the utmost good faith degree, and the more meeting of this method needle track causes other complication in addition.Ultrasonic can the correction in the art is shifted, and the error of navigation system is reduced, but ultrasonoscopy is difficult to the following nuclear of identification 1cm group, also is difficult to differentiate acoustic impedance and differs very little alba and ectocinerea.

Summary of the invention

Technical problem to be solved by this invention is to provide a kind of enforcement simple at the defective that prior art exists, flexible operation, the real-time height, human body there is not injury, and need not to change the method for the surgical navigation systems accuracy correction that has navigator now, be specially a kind of method for real-time error correction of neurosurgery navigation puncture path based near infrared spectrum.

The present invention adopts following technical scheme for achieving the above object:

The present invention is based on the method for real-time error correction of neurosurgery navigation puncture path of near infrared spectrum, it is characterized in that comprising the steps:

(1) the MRI data is obtained the three-dimensional reconstruction body through the volume drawing accelerating algorithm three-dimensional reconstruction based on ray cast;

(2) adopt the parametric equation of the straight line algorithm from the volumetric data set of the described three-dimensional reconstruction body of step (1), to extract destination path;

(3) resample points when the described destination path of step (2) is the voxel that exists in the volumetric data set, then the half-tone information of the destination path actual value R that promptly punctures _MRIDraw according to the location index of this voxel in volumetric data set; The actual value R otherwise the half-tone information that adopts the Tri linear interpolation algorithm to obtain the described destination path of step (2) promptly punctures _MRI

(4) adopt the cone method that the described destination path of step (2) is carried out circular coneization and obtain N cone, the path half-tone information that adopts the Tri linear interpolation algorithm to extract the taper seat of a described N cone respectively obtains N bar checking puncture track, and obtain the half-tone information that the N bar is verified track, wherein N is the natural number less than 90;

(5) record puncture track near-infrared parameter D under the state that does not have mark _FNIRS, IR parameters D nearly _FNIRSThrough image greyscale information and near-infrared parameter D _FNIRSBetween the mathematical model conversation track data D that obtains puncturing _MRI

(6) as the described puncture track data of step (5) D _MRIWith the described puncture actual value of step (3) R _MRIBetween matching error less than the range of error of setting, then finish error correction;

(7) as the described puncture track data of step (5) D _MRIWith the described puncture actual value of step (3) R _MRIBetween matching error greater than the range of error of setting, then adopt curvilinear trend matching algorithm based on Hausdorff distance and curvature with the described puncture track data of step (5) D _MRIMate the track data D that obtains and puncture with the half-tone information of the described N bar checking of step (4) track _MRIThe checking track that variation tendency is identical, and with described and puncture track data D _MRIThe checking track that variation tendency is identical is adjusted puncture track data D in real time through the three dimensional space coordinate drift rate _MRI, with adjusted puncture track data D _MRIWith the described puncture actual value of step (3) R _MRIAgain mate.

The present invention combines fMRI (containing MRI information and BOLD information) with function near-infrared spectrum technique (fNIRS), set up the optimum mathematics model of neurosurgery navigation and target spot identification, reaches the purpose of real-time navigation, target spot identification and real-time error.Near-infrared spectral analysis technology is as the measurement means of a kind of, noinvasive harmless to human body, no ionizing radiation, its optical parametric and medical image data exist close related, for in real time cerebral tissue characteristic, the identification target spot in acquisition probe the place ahead provide reliable foundation, for the identification of the real-time tracking error correction in path in the neurosurgery navigation and target position provides may.Routing information extracts and adopts the parametric equation of the straight line algorithm, simultaneously in order to find the coupling routing information quickly and accurately, adopt the Hausdorff distance and the curvature algorithm obtains and the match information of the track that punctures, the orientation that near-infrared optical signal L departs from objectives is determined in orientation by the checking curve place that obtains, the coordinates computed skew is carried out the path and is proofreaied and correct.Matching algorithm is combined with the destination path half-tone information, can effectively carry out real-time correction the puncture course deviation in the art.The inventive method is implemented simple, and flexible operation need not to change existing navigator, is convenient to clinical practice.

Description of drawings

Fig. 1 is the principle schematic of near-infrared monitor on the throne.

Fig. 2 is the near infrared spectrum slope curve of department of cerebral surgery navigation system and the corresponding relation of MRI image, and explanation can adopt the slope method of near infrared light mathematic(al) parameter tentatively to finish navigation task.

Fig. 3 is that optical parametric, blood oxygen parameter and the hemodynamic parameter of 5 rat brain cortex distributes.Fig. 3 (a) and Fig. 3 (b) are respectively under 5 rat brain cortex skulls the change along with the degree of depth, and it optimizes the change curve of scattering coefficient and absorptance (wavelength 834nm), and NO.1 is the 1st mouse, and NO.2 is the 2nd mouse ..., and the like; Fig. 3 (c) and Fig. 3 (d) are respectively under 5 rat brain cortex skulls the change along with the degree of depth, the change curve of its total hemoglobin concentration and blood oxygen saturation; Fig. 3 (e) and Fig. 3 (f) are respectively under 5 rat brain cortex skulls the change along with the degree of depth, the change curve of its blood flow and blood volume.The measured every index that obtains all is in the coverage of data that document provides, and fathoms a little at each, and the fluctuation of measured parameter value is very limited, provides foundation for realizing that optics is dissected the location in the brain.

Fig. 4 navigation sketch map that to be near infrared spectrum (fNIRS) merge mutually with the MRI image data.Fig. 4 (a) carries out the anatomical structure figure of data fusion for fNIRs and MRI, wherein A is the operation puncturing point, B is the operation target spot, selected image is the fusion results of MRI and fMRI image, information by MRI gradation of image data fetch puncture path, by the BOLD information (blood oxygen information) of fMRI, on puncture path, avoid the critical function district.Fig. 4 (b) is a fNIRs puncture path error correction sketch map, and intermediary dotted line is actual fNIRs puncture track, and the conical region of AB～AC is the possible offset area of navigation puncture.

Fig. 5 is the puncture path planning chart based on the near infrared spectrum navigation.Wherein, Fig. 5 (a) is the path planning figure based on said three-dimensional body, the summit of taper is for going into to sting a little, path along taper surface is a path planning, obviously path planning is many more, and matching precision is high more, because the limitation that shows, only provide limited several path plannings among the figure, each path planning has the offset information of top to bottom, left and right, front and rear simultaneously; Fig. 5 (b) is the half-tone information of actual puncture path; Fig. 5 (c) is one of possibility offset path information (intended path) in the operation process; Fig. 5 (d) is the contrast of actual puncture path and possibility offset path, by route matching, can obtain the angle and direction that offset path departs from Actual path, is used to guide the error correction of operation puncturing process.

Fig. 6 is a flow chart of the present invention.

The specific embodiment

Be elaborated below in conjunction with the technical scheme of accompanying drawing to invention:

As shown in Figure 1, be the principle schematic of near-infrared monitor on the throne.The incident source is the radio-frequency signal source (RF Source) of multi-wavelength, near-infrared laser after frequency divider (Spliter) modulation projects biological tissue (Tissue) by sending optical fiber (Delivery fiber), each road light source is according to logical sequence (Switch) work of setting, and LD is a laser diode; Has stronger tissue penetration ability by the what near infrared light, this light enters tissue, after the absorption and scattering through tissue, after accepting the optical fiber collection and being sent to photomultiplier tube (PMT), process amplifier (Amplifier), band filter (BP Filter), demodulator (Demodulator), low pass filter devices such as (LP Filter), the output signal of telecommunication converts digital signal to by analog-digital converter (ADC), by computer data acquiring and processing, measured signal can be shown in computer screen in real time.This circuit design has adopted the high frequency modulated demodulation techniques, the amplitude of optical signal and position is interrelated with optical properties of tissue (absorbing and scattering coefficient), thereby can extrapolate the biological nature (blood oxygen, blood volume, blood flow etc.) of tissue.

As shown in Figure 2, be the near infrared spectrum slope curve of department of cerebral surgery navigation system and the corresponding relation of MRI image, explanation can adopt the slope method of near infrared light mathematic(al) parameter tentatively to finish navigation task.

As shown in Figure 6, complete data fusion and guidance path real-time error process are: scan-data before the MRI art → extraction puncture track MRI gradation of image data, for the bias correcting of guided puncture track under the situation of systematic error can enlarge taper (N bar plan puncture track, probable value as puncture planning) → in real time fNIRs image data → fNIRs puncture track real time data measured value merges the probable value coupling with N bar plan track) → data fusion result's (goodness of fit of calculating fNIRs track and image path, find out the probable value that mates most on the statistical significance, and the real-time error of the three dimensional space coordinate deviant guiding operation pathway stored of image application data) → and draw the piercing process handling suggestion, carry out the real-time three-dimensional error correction.

The present invention realizes through the following steps:

Volume drawing accelerating algorithm based on ray cast

The extraction of going into thorn point and target spot routing information before the operation is the foundation of operation pathway real-time error, and the three-dimensional reconstruction of MRI sequence image is the key that the volume data routing information extracts.The present invention adopts a kind of ray cast volume drawing accelerating algorithm that merges based on fragment, algorithm utilizes throw light and bundle of planes to ask friendship, determine to merge fragment fast, employing is based on segmental fusion rendering technique, accelerate fusion speed, and utilize the bounding box technology to reduce, improved the efficient of ray cast the invalid planar friendship of asking.Because the pixel in the fragment has similar optical properties, based on the data consistency analysis to volume data, by the drafting equation based on pixel, can derive obtains based on segmental drafting equation, with the method for iteration, draw equation according to the past order backward and be expressed as:

$C_{\mathrm{out}}=C_{\mathrm{now}}(1-{\mathrm{\α}}_{\mathrm{in}})\underset{k=0}{\overset{n_i-1}{\mathrm{\Σ}}}{(1-{\mathrm{\α}}_{\mathrm{now}})}^k+C_{\mathrm{in}}$

${\mathrm{\α}}_{\mathrm{out}}={\mathrm{\α}}_{\mathrm{now}}(1-{\mathrm{\α}}_{\mathrm{in}})\underset{k=0}{\overset{n_i-1}{\mathrm{\Σ}}}{(1-{\mathrm{\α}}_{\mathrm{now}})}^k+{\mathrm{\α}}_{\mathrm{in}}$

C wherein _NowAnd α _NowBe the optical properties (color value and opacity value) of last intersection point of fragment, C _InAnd α _InBe the optical properties of segmental initial intersection point, C _OutAnd α _OutBe the optical properties of the resample points between two intersection points of fragment, n _iBe segmental length.

Extract the path half-tone information

Go into a thorn point A (x known ₀, y ₀, z ₀) and target spot B (x ₁, y ₁, z ₁) situation under, can adopt the straight line parametric technique to give expression to the equation expression formula in operation puncturing path: X=x ₀+ mt, Y=y ₀+ nt, Z=z ₀+ pt, wherein, l={m, n, p} are the direction vector of linear equation, m, n, p be l respectively at x, y, the component on three coordinate axess of z, t are arbitrary parameter and are not 0, x ₀, y ₀, z ₀Be respectively (x into thorn point A ₀, y ₀, z ₀) at x, y, the coordinate figure of three directions of z, X, Y, Z is target spot being had a few to the linear equation of going into thorn point.Adopt the method that resamples to extract this puncture path half-tone information then, if the voxel that exists in the resample points volumetric data set of puncture path L, its half-tone information draws according to the location index of this voxel in volumetric data set, and the gray value of the resample points that can't index obtains is obtained by the Tri linear interpolation algorithm.Under the situation of the initial value of determining resampling parameter and volume elements numbering, can determine the volume elements numbering of all the other resample points parameters and throw light process with following recurrence relation:

If: $d_x^{k+1}=d_x^k+{\mathrm{\δ}}_x>l$ And δ V is along the x direction of principal axis, then: $d_x^{k+1}\←d_x^{k+1}-l,$ i←i+1；

If: $d_x^{k+1}=d_x^k+{\mathrm{\δ}}_x>l$ And δ V is along the x axle in the other direction, then: $d_x^{k+1}\←d_x^{k+1}-l,$ i←i-1；

If: $d_x^{k+1}=d_x^k+{\mathrm{\δ}}_x\≤l,$ Then: d _x ^K+1Constant, i is constant;

Wherein volume elements be numbered (m), i, j, m are respectively central point x in three-dimensional coordinate of volume elements for i, j, y, the coordinate figure of three directions of z, the length of volume elements is respectively l, w, h; The sampling interval vector is δ V, and (l, w h) are sampling step length to δ≤min, and δ x, δ y, δ z are sampling interval vector δ V decomposition amount along three coordinate axess in object coordinates system, d _x, d _y, d _zBe the distance of the outer surface of sampled point, d along three change in coordinate axis direction to volume elements _x ^kRepresent the outer surface distance of this sampled point, ← expression assignment relation, the d after the assignment along the x direction of principal axis to the place volume elements _x ^K+1Represent the distance of next sampled point along the x axle to own place volume elements outer surface; Axial parameter can the rest may be inferred along y and z for resample points.

F (i, j, m) be volume elements (i, j, gray value m), this value is obtained by the Tri linear interpolation algorithm.Algorithm is: (m) data of 8 nearest consecutive points are respectively: f for i, j to establish resample points ₀₀₀, f ₀₀₁, f ₀₁₀, f ₀₁₁, f ₁₀₀, f ₁₀₁, f ₁₁₁, f ₁₁₀, d _x, d _y, d _zRepresent sampled point respectively with respect to 000 o'clock distance at three coordinate directions, then the value f of this sampled point (i, j m) are:

f(i，j，m)＝(1-d _x)×(1-d _y)×(1-d _z)×f ₀₀₀+d _x×(1-d _y)×(1-d _z)×f ₀₀₁

+(1-d _x)×d _y×(1-d _z)×f ₀₁₀+d _x×d _y×(1-d _z)×f ₀₁₁+。

(1-d _x)×(1-d _y)×d _z×f ₁₀₀+d _x×(1-d _y)×d _z×f ₁₀₁+

(1-d _x)×d _y×d _z×f ₁₁₀+d _x×d _y×d _z×f ₁₁₁

The mathematical model of view data and near-infrared optical data fusion

Newest research results shows, the result of near-infrared test result and medical image exists close related, the near-infrared measuring result has same reliability to the reflection of biological tissue's information and the result of MRI, and the real-time of near-infrared measuring parameter and " observability " can remedy the limitation of image navigation system applies.The near infrared absorption coefficient, the anatomic information of distribution of scattering coefficient tissue morphology and MRI image is corresponding substantially, the anatomic information of MRI image has reflected the H cuclear density distribution (corresponding tissue density distributes) of different cerebral tissue, the near infrared light mathematic(al) parameter has reflected organizes elastic photon scattering coefficient distribution (also distributing relevant with tissue density), obviously, if the mathematical model that can find MRI data and near infrared light mathematic(al) parameter to merge, but by just accurate recording probe the place ahead cerebral tissue Density Distribution of near-infrared parameter in the art, thereby guiding operation pathway and real-time error further improve positioning accuracy.

Concrete grammar is to utilize the rat brain stereotactic apparatus, with near-infrared Wicresoft continuous monitor system on the throne and special Wicresoft probe the optical parametric of a plurality of tracks of rat cerebral tissue is tested, again rat is carried out MRI scanning, obtain the MRI data on the near-infrared needle track track.The pseudo-shadow that carries out optical parametric after MRI data and the near infrared light mathematic(al) parameter normalized is again eliminated, sought the dependency between the two, and set up the dependency mathematical model.

Curvilinear trend matching algorithm based on Hausdorff distance and curvature

Employing guides the real-time error of the puncture path in the operation process based on the curvilinear trend matching algorithm of Hausdorff distance and curvature.For in operation process, provide the deviation angle and the offset direction of puncture path in real time, it is the near-infrared optical signal that thereby guiding in real time is established any one section curve L, L and grey scale curve are carried out the trend contrast, if in certain tolerance ε scope, then near-infrared puncture track L and destination path match; If surpass tolerance ε, then L and N bar checking curve are carried out the variation tendency contrast, obtain in tolerance ε scope, to have the checking puncture track N ' of identical change trend with curve L.Its mean curvature algorithm is:

If curvilinear equation is s=s (t), t is a parameter of curve, and the computing formula of curvature is:

$k\left(t\right)=\frac{|s^{\′}\left(t\right)\×s^{\′\′}\left(t\right)|}{{\left[s^{\′}\left(t\right)\right]}^3},$ Wherein s ' (t), s " (t) is first derivative and the second dervative of curvilinear equation s (t).The Hausdorff distance definition is: be provided with two groups of finite aggregate A={a ₁, a ₂..., a _pAnd B={b ₁, b ₂..., b _p), wherein, a ₁, a ₂..., a _pBe all elements of set A, b ₁, b ₂..., b _pBe all elements of set B, then the Hausdorff distance definition is between A, B:

H (A, B)=max (max a _s∈ Amin b _t∈ B||a _s-b _t||, max b _t∈ Bmin a _s∈ A||b _t-a _s||), in the formula, || ... || represent certain definition apart from normal form, two directed distances that are called A-B and B-A in the bracket.The Hausdorff distance metric be two maximums between set degree that do not match, and calculate easyly, therefore between points one-to-one relationship in also overcritical two set is suitable for the coupling of image very much.If the set of the curvature of two curves is C ₁={ K _1s| s=1,2 ... m), C ₂={ K _2t| t=1,2 ... n}, wherein K _1sAnd K _2tBe respectively the curvature value at each place, summit behind two curve polygonal approximations, m and n are any positive integer here, and m and n not necessarily equate.When route matching, press the Hausdorff distance that following formula calculates two curvature collection, it is right that its intermediate value minimum a pair of promptly corresponds to the profile that may mate.Here be the simple comparison of curvature value apart from normal form, can improve the efficient of algorithm greatly, be verified curve.

The orientation that near-infrared optical signal L departs from objectives is determined in orientation by the checking curve place that obtains, and the coordinates computed skew is carried out the path and proofreaied and correct.

The present invention introduces the curvilinear trend matching algorithm based on Hausdorff distance and curvature, in conjunction with the half-tone information that extracts, both guaranteed the precision of prediction of path offset, solve the path again and proofreaied and correct this difficult problem, thereby can implement clinically, increase substantially the precision of surgical navigational.

Embodiment

3 to Fig. 5 to narrate the invention process as follows in conjunction with the accompanying drawings:

As shown in Figure 3.The optical parametric, blood oxygen parameter and the hemodynamic parameter that are 5 rat brain cortex distribute.Fig. 3 (a) and Fig. 3 (b) are respectively under 5 rat brain cortex skulls the change along with the degree of depth, and it optimizes the change curve of scattering coefficient and absorptance (wavelength 834nm), and NO.1 is the 1st mouse, and NO.2 is the 2nd mouse ..., and the like; Fig. 3 (c) and Fig. 3 (d) are respectively under 5 rat brain cortex skulls the change along with the degree of depth, the change curve of its total hemoglobin concentration and blood oxygen saturation; , Fig. 3 (e) and Fig. 3 (f) are respectively under 5 rat brain cortex skulls the change along with the degree of depth, the change curve of its blood flow and blood volume.The measured every index that obtains all is in the coverage of data that document provides, and fathoms a little at each, and the fluctuation of measured parameter value is very limited, provides foundation for realizing that optics is dissected the location in the brain.

1. at first carry out the fusion treatment of rat function near infrared spectrum (fNIRS) parameter and MRI image data, set up related mathematical model.Utilize fNIRS near-infrared Wicresoft continuous monitor system on the throne and the special Wicresoft probe developed, gather rat cerebral tissue and measure each point optical parametric (ScO on the puncture track ₂, μ a, μ s, Hb and HbO ₂Concentration), obtain the near-infrared data D of surgical planning puncture track _FNIRSAlong the puncture course bearing rat is carried out the MRI image scan simultaneously, obtain the MRI grey scale signal D on the near-infrared needle track track _MRI, with D _MRIAnd D _FNIRSData normalization is handled the two dependency of post analysis, seeks the mathematical model between the two.When setting up path association mathematical model, experiment is intended adopting and is designed the brain solid positioning framework that can put into NMR system, fixes animal, and positions mark.

As shown in Figure 4, the navigation sketch map that merges mutually near infrared spectrum (fNIRS) and MRI image data.Fig. 4 (a) carries out the anatomical structure figure of data fusion for fNIRs and MRI, wherein A is the operation puncturing point, B is the operation target spot, selected image is the fusion results of MRI and fMRI image, information by MRI gradation of image data fetch puncture path, by the BOLD information (blood oxygen information) of fMRI, on puncture path, avoid the critical function district.Fig. 4 (b) is a fNIRs puncture path error correction sketch map, and intermediary dotted line is actual fNIRs puncture track, and the conical region of AB～AC is the possible offset area of navigation puncture.

2. adopt the Fast Volume Rendering Algorithm algorithm that merges based on segment at 256 * 256 * 84 sequence MRI image, obtain the three-dimensional reconstruction body of cerebral tissue data, the three-dimensional reconstruction body is carried out the translation of arbitrarily angled rotation, any direction; In order to obtain more real display effect, the parameter that threedimensional model is made amendment interface is provided, comprising: object color, background color, material, diffuse-reflectance, Ambient, and light selection remove sawtooth effect etc.; In said three-dimensional body, can choose the target area, measure, cut and separate, rebuild human tissue organ's perspective view from the puncture direction.The BOLD information that on the three-dimensional reconstruction body, reflects the critical function district by the brain function analysis software.

As shown in Figure 5, be puncture path planning chart based on the near infrared spectrum navigation.Wherein, Fig. 5 (a) is the path planning figure based on said three-dimensional body, the summit of taper is for going into to sting a little, path along taper surface is a path planning, obviously path planning is many more, and matching precision is high more, because the limitation that shows, only provide limited several path plannings among the figure, each path planning has the offset information of top to bottom, left and right, front and rear simultaneously; Fig. 5 (b) is the half-tone information of actual puncture path; Fig. 5 (c) is one of possibility offset path information (intended path) in the operation process; Fig. 5 (d) is the contrast of actual puncture path and possibility offset path, by route matching, can obtain the angle and direction that offset path departs from Actual path, is used to guide the error correction of operation puncturing process.

3. on the three-dimensional reconstruction body, extract planning puncture path (select convenient, the safest operative approach, avoid functional areas) and seek optimum puncturing point, reduce functional lesion when farthest damaging focus, increase operation safety.Utilize the point of 8 neighborhoods to adopt the Tri linear interpolation algorithm to extract half-tone information on the puncture path.With the image information of the best puncture track of extracting as puncture actual value R _MRI, and be that central shaft carries out taper and enlarges with this path, extract N bar plan puncture track as possible deviation path, obtain the three dimensional space coordinate drift rate (deviation angle and orientation) of probable value and actual value.

4. under the situation of not having the location mark, write down puncture track fNIRS parameter D _FNIRS, pass through D _MRIAnd D _FNIRSBetween the mathematical model, IR parameters D nearly _FNIRSBe converted into the half-tone information D of image _MRI, the D that adopts the curvilinear trend matching algorithm based on Hausdorff distance and curvature that conversion is obtained _MRIInformation and puncture actual value R _MRIMate with N bar plan puncture trace image data, carry out real-time small adjustment according to three dimensional space coordinate drift rate (deviation angle and orientation) simultaneously, up to the puncture track data D that is obtained _MRI(by real-time D _FNIRSData conversion obtains) and puncture actual value R _MRIBetween matching error within the systematic error scope, just finish the real-time route error correction of surgical navigational process.

Claims (5)

1. A real-time error correction method for neurosurgery navigation puncture path based on near-infrared spectrum, characterized in that it comprises the following steps: (1) 3D reconstruction of MRI data through a volume rendering acceleration algorithm based on ray projection to obtain a 3D reconstruction volume; (2) adopting the linear parametric equation algorithm to extract the target path from the volume data set of the three-dimensional reconstruction body described in step (1); (3) When the resampling point of the target path described in step (2) is a voxel in the volume data set, then the gray information of the target path is the actual value of puncture R _MRI according to the position index of the voxel in the volume data set Draw; Otherwise, adopt the trilinear interpolation algorithm to obtain the gray information of the target path described in step (2) that is the puncture actual value R _MRI ; (4) Cone method is used to conize the target path described in step (2) to obtain N cones, and a trilinear interpolation algorithm is used to extract the path gray information of the conical surfaces of the N cones respectively to obtain N pieces Verify the puncture track and obtain the grayscale information of N verification tracks, where N is a natural number less than 90; (5) Record the near-infrared parameter D _fNIRS of the puncture trajectory in the state without marking, and convert the near-infrared parameter D _fNIRS through the mathematical correlation model between the image grayscale information and the near-infrared parameter D _fNIRS to obtain the puncture trajectory data D _MRI ; (6) When the matching error between the puncture trajectory data D _MRI described in step (5) and the actual puncture value R _MRI described in step (3) is smaller than the set error range, then the error correction is completed; (7) When the matching error between the puncture trajectory data D _MRI described in step (5) and the actual puncture value R _MRI described in step (3) is greater than the set error range, the curve based on Hausdorff distance and curvature is used The trend matching algorithm matches the puncture trajectory data D _MRI described in step (5) with the gray information of the N verification trajectories described in step (4) to obtain a verification trajectory with the same variation trend as the puncture trajectory data D _MRI , and The verification trajectory with the same variation trend as _{the puncture trajectory data D MRI is adjusted in real time through the offset degree of the three-dimensional space coordinates, and the puncture trajectory data D MRI is adjusted} in real time. _MRI rematched. 2. The real-time error correction method for neurosurgery navigation puncture path based on near-infrared spectrum according to claim 1, characterized in that the volume rendering acceleration algorithm based on ray projection described in step (1) is: MRI data from the imaging plane Each pixel of the system emits a ray according to the set observation direction, and the ray passes through the voxel matrix of the three-dimensional data field, and then sets the sampling space through the bounding box method and the fragment-based fusion rendering method, and the ray in the sampling space Resampling is performed to obtain the opacity and color values of all sampling points on the ray, and the opacity and color values are synthesized from front to back based on the light absorption and emission model for the resampling points to obtain the volume data set of the 3D reconstruction volume: ${\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; out</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; now</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; in</span>}}<span\; class="notranslate">\; )</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; now</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; in</span>}}<span\; class="notranslate">\; ,</span>$ ${\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; out</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; now</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; in</span>}}<span\; class="notranslate">\; )</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; now</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; in</span>}}$ Among them, C _now is the color value of the last intersection point of the MRI data segment, α _now is the opacity value of the last intersection point of the MRI data segment, C _in is the color value of the initial intersection point of the MRI data segment, and α _in is the starting point of the MRI data segment The opacity value of the initial intersection point, C _out the color value of the resampling point between two intersection points of the MRI data segment, α _out is the opacity value of the resampling point between two intersection points of the MRI data segment, n _i is the segment's length, k is a weighted value, and both _ni and k are positive integers. 3. The near-infrared spectrum-based neurosurgery navigation puncture path real-time error correction method according to claim 1, characterized in that the linear parameter equation algorithm described in step (2) is: the known puncture point A(x ₀ , y ₀ , z ₀ ) and target point B(x ₁ , y ₁ , z ₁ ) get the direction vector l={m,n,p} between the pricking point and the target point, then the parametric equation of the line is: $\left\{\begin{array}{c}\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; mt</span>}\\ \mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; nt</span>}\\ \mathrm{<span\; class="notranslate">\; Z</span>}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; z</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; pt</span>}\end{array}\right.<span\; class="notranslate">\; ,</span>$ Where t is an arbitrary parameter and not 0, x ₀ , y ₀ , z ₀ are the values of the pricking point A (x ₀ , y ₀ , z ₀ ) on the x, y, z three-dimensional coordinate axes, x ₁ , y ₁ , z ₁ are the components of the target point B (x ₁ , y ₁ , z ₁ ) on the x, y, z three-dimensional coordinate axes respectively, m, n, p are the direction vectors l respectively on x, y, z Components on the three-dimensional coordinate axes, points X, Y, and Z of the parametric equation of the straight line connecting the target point to the pricking point get the target path. 4. The near-infrared spectrum-based neurosurgical navigation puncture path real-time error correction method according to claim 1, characterized in that the trilinear interpolation algorithm described in step (3) and step (4) is as follows: Set the initial values of the resampling parameters and voxel number, determine the remaining resampling point parameters and the voxel number that the projection ray passes through, and obtain the resampling point along the x, y, z axis direction parameters and the voxel number that the projection ray passes through The method is the same, where the parameters of the resampling point along the x-axis direction are obtained as follows: when $d_x^{k+1}=d_x^k+{\mathrm{\&delta;}}_x>l$ And δ·V is along the x-axis direction, then: update d _x ^k+1 to d _x ^k +δ _x -l, i+1; when $d_x^{k+1}=d_x^k+{\mathrm{\&delta;}}_x>l$ And δ·V is along the opposite direction of the x-axis, then: update d _x ^k+1 to d _x ^k +δ _x -l, i-1; when $d_x^{k+1}=d_x^k+{\mathrm{\&delta;}}_x\&le;l,$ Then: d _x ^k+1 remains unchanged, i remains unchanged; The voxel numbers are (i, j, m), i, j, and m are the coordinate values of the center point of the voxel in the three-dimensional coordinates of x, y, and z respectively, and the length, width, and height of the voxel are respectively l, w, h, δ V is the sampling interval vector, δ≤min(l, w, h) is the sampling step size, δ _x , δ _y , δ _z is the sampling interval vector δ V in the object coordinate system The amount of decomposition along the three coordinate axes, d _x ^k represents the distance from the sampling point, that is, the kth point along the x-axis direction, to the outer surface of the voxel where it is located, and the updated d _x ^k+1 represents the next sampling point, namely k+1 The distance from the point along the x-axis to the outer surface of the voxel where it is located; The gray value f(i, j, m) of the target path is obtained by using the parameters of the resampling point along the x, y, and z axes and the voxel number that the projected light passes through: the resampling point is the voxel (i, j, m ) the nearest 8 adjacent point data, that is, the first point data f ₀₀₀ to the eighth point data f ₁₁₁ , d _x , d _y , d _z represent the distances of the resampling point relative to the first point in three coordinate directions, Then the gray value f(i, j, m) of the resampling point is: f(i,j,m)=(1-d _x )×(1-d _y )×(1-d _z )×f ₀₀₀ +d _x ×(1-d _y )×(1-d _z )× f ₀₀₁ +(1-d _x )×d _y ×(1-d _z )×f ₀₁₀ +d _x ×d _y ×(1-d _z )×f ₀₁₁ + All resampled (1-d _x )×(1-d _y )×d _z ×f ₁₀₀ +d _x ×(1-d _y )×d _z ×f ₁₀₁ + (1-d _x )×d _y ×d _z ×f ₁₁₀ +d _x ×d _y ×d _z ×f ₁₁₁ The gray value of the point is the gray information that constitutes the target path. 5. The near-infrared spectrum-based neurosurgery navigation puncture path real-time error correction method according to claim 1, characterized in that the method of coning the target path in step (4) is: taking the target path as the vertical line , with the piercing point as the apex of the cone, and the angle of the apex of the cone varies continuously from 1 to N degrees at an interval of 1 degree to draw a cone to obtain N conical regions, where N is a natural number less than 90.

Images